FPGA Central

I've been reading about symbol recovery loops using the Gardner and Mueller
algorithms. All of the examples I have read seem clear, but the examples
all use QPSK or BPSK. Also, I notice that Gardner's paper (which I haven't
read) is titled "A BPSK/QPSK Timing Error Detector for Sampled Receivers".
Do either of these algorithms apply to QAM? It isn't clear to me that they
do.
-Kevin

On Mon, 26 Jan 2004 20:22:19 GMT, "Kevin Neilson"
<[email protected]> wrote:

>I've been reading about symbol recovery loops using the Gardner and Mueller
>algorithms. All of the examples I have read seem clear, but the examples
>all use QPSK or BPSK. Also, I notice that Gardner's paper (which I haven't
>read) is titled "A BPSK/QPSK Timing Error Detector for Sampled Receivers".
>Do either of these algorithms apply to QAM? It isn't clear to me that they
>do.
>-Kevin
>

Yes they actually do. If you look at the calculations in the
algorithms carefully, you can just see that they are using the
amplitudes of the sampled data in baseband. So BPSK is just a two
level signal, QPSK is a quadrature signal in baseband. M&M also can be
used for two level signals or two level quadrature signals. When you
go to PAM in baseband, signal changes become more difficult to
differentiate (is a change from -3 to -1 the same as -1 to 3?) so you
may have to train the symbol timing (this is what you mean, right)
loop with a two level signal first.

Muzaffer Kal

http://www.dspia.com
ASIC/FPGA design/verification consulting specializing in DSP algorithm implementations